At present, large calculation amount is still a bottleneck restricting the development of the robot real-time positioning technology. In the existing robot real-time positioning method, usually, the robot pre-stores the environment map, acquires external environmental parameters, and calculates its current position in the map based on the acquired parameters; or the robot constructs the environment map based on the acquired parameters and then calculates its current location in the map, so as to positioning the robot itself. Typically, the amount of calculation is large, which places high requirements on the hardware and software of the robot. In addition, a large amount of calculation is disadvantageous to the fast navigation for the robot.